|
Part I
The Mechanism
|
Silver. Down 53% from its January all-time high. Bloomberg says the solar trade is over — photovoltaic silver demand fell 19% this year. CNBC ran a segment titled "Silver's Solar Story Fades." If you've been reading this letter, you already know the next move: they're staring at the dashboard. The engine is somewhere else entirely.
Here's what nobody on the sell side has bothered to diagram: 73.9% of the world's mined silver is a byproduct. Not a co-product. Not a secondary focus. A rounding error in someone else's cost report. Lead miners, copper miners, zinc miners — they pull silver out of the ground because it happens to be sitting in the same rock. They don't increase output when the price doubles. They don't decrease it when it halves. The silver simply comes or doesn't, and nobody at the mine is making that decision.
The price rose 145% in 2025. Mine output rose 3%. That's not a lag — that's the geology telling you the supply curve is someone else's business plan. The quarter of output that actually can respond — primary silver mines — shrank. For the third consecutive year. Fresnillo, the world's largest primary producer, cut output 13.5% and guided 2026 lower. Not because of the price. Because the stopes are depleted. The rock doesn't care about your chart.
I traded silver on the solar thesis in early 2025. It worked beautifully until the thrifting numbers hit. What I missed then — and what most of the market is still missing — is that silver's demand base has quietly migrated out from under the one story everyone is watching.
|
Part II
The Diagram
|
Story off. Numbers on.
Total demand: 1.10 billion ounces. Total mine supply: 844.1 million ounces. Recycling adds 211.3 million. Still not enough. The difference gets drawn from above-ground stockpiles — vaults, ETFs, central bank reserves, industrial buffers. Roughly 800 million ounces have been pulled from those reserves since 2021. That reservoir is not infinite.
Where the metal actually comes from tells the whole story. Lead and zinc mines produced 249.1 Moz — 29.4% of global supply. Copper mines: 237.3 Moz, 28.0%. Primary silver mines: 221.1 Moz, 26.1% — a record low share. Gold mines: 134.7 Moz. Six of the top ten silver-producing companies on Earth are primarily digging for base metals.
Now here's where the bears got sloppy. Solar thrifting is real — PV silver demand fell 19% this year through thinner paste layers and copper-coated substitutes. But the demand didn't vanish. It migrated. The Silver Institute's electrical and electronics category — AI data centers, 5G infrastructure, EV power systems — now runs at 422.9 million ounces annually. AI data center demand alone is expected to reach roughly 42.3 million ounces by 2027. That nearly replaces what solar shed. Nobody is running that headline.
Mine lead times average 15.7 years from discovery to production. Escobal, a 20-million-ounce-per-year mine in Guatemala, has sat idle since 2017 over an indigenous consultation ruling. There is no supply cavalry. The pipeline is empty and the permitting queue stretches to 2040.
|
Part III
The Weak Link
|
Two pressure points. One is in New York. The other is in Beijing. Neither is visible on a price chart.
COMEX registered silver — the metal actually deliverable against futures contracts — sits at 103.1 million ounces as of October 13. Open interest represents roughly 567 million ounces of exposure. That's a coverage ratio of 18.2%. For every ounce of deliverable silver in the vault, there are roughly five paper claims on it.
This isn't unusual by COMEX standards — the system was built on trust, not full backing. But at 18.2%, a single large delivery demand could reset the whole structure. I've watched this movie before with palladium. The coverage looked fine right up until it didn't.
Then there's Beijing. China replaced its old export quota system with a licensing framework back in 2019, and has steadily tightened the regime since. Only 44 state-approved companies can now export silver. Minimum requirements include 80 tonnes annual production. About 120 million ounces of annual silver exports now flow through a government-controlled gate that can narrow without warning.
China's imports tell the real story. March 2026 silver imports surged 173% above the 10-year seasonal average — the highest on record. In the first two months alone, China pulled in over 790 tonnes. They're not exporting generously. They're accumulating. I saw the same playbook with gallium. Beijing doesn't ban outright — that triggers WTO complaints. They just add friction until the gate functionally closes.
The industry reported all-in sustaining costs of $12.21 an ounce in 2025. That figure is a fiction built on byproduct credits — the only pure silver mine in the dataset, Aya Gold & Silver's Zgounder in Morocco, runs a cash cost of $20.41. At $57 spot, even the expensive mines print money. The price signal to produce more is screaming. The geology and the permitting timelines cannot hear it.
|
Part IV
The Chain Reaction
|
The sequence is mechanical. If you've watched physical metals during an inventory drain, the choreography doesn't change — only the timing.
Deficit year six rolls into year seven. Above-ground stocks, already down roughly 800 million ounces since 2021, keep draining. At some point — and I don't pretend to know whether it's this quarter or next year — somebody large takes delivery rather than rolling their futures contract. That's the crack. COMEX registered draws down hard, coverage drops below 10%, and the LBMA-COMEX spread blows out.
Then the physical buyers who were sitting on the sideline — the ones waiting for premiums to cool — realize the metal they need isn't sitting in a vault anymore. They all hit the bid at once. Third wave: paper shorts cover. Silver did something like this in 2020–2021. The Reddit squeeze wasn't the cause. It was just the match. The deficit was already built. It's built again, wider this time.
Where does the capital go? Not into SLV. That ETF holds roughly 500 million ounces of physical silver and has become part of the plumbing that buffers the deficit. Large ETF redemptions don't fix the problem — they accelerate it. The edge, if there is one, is in primary silver producers with unhedged spot exposure and actual operating mines. Companies whose revenue is denominated in spot silver, not streaming contracts locked in at $5 an ounce years ago. Hecla. Pan American's silver segment. Coeur Mining. The ones whose costs are known and whose leverage to spot is direct.
The sell side says the solar story is over. The physical market says the deficit is getting worse. I know which one I've learned to trust. It just takes longer than you want it to.
Sources: World Silver Survey 2026 (Silver Institute / Metals Focus), CME Group COMEX Warehouse Reports, Chinese Customs Authority, S&P Global Market Intelligence, Fresnillo plc Preliminary Results 2025, Aya Gold & Silver Q4 2025 Results
